Guidance on Uncertainty Assessment - European Commission ...

Template No. 4: Annual emissions report of stationary installationsTemplate No. 5: Annual emissions report of aircraft operatorsTemplate No. 6: Tonne-kilometre data report of aircraft operatorsBesides these documents dedicated to the MRR, a separate set of guidance documentson the A&V Regulation is available under the same address.All EU legislation is found on EUR-Lex: http://eur-lex.europa.eu/The most important legislation is furthermore listed in the Annex of this document.Also competent authorities in the Member States may provide useful guidance on theirown websites. Operators of installations should in particular check if the competentauthority provides workshops, FAQs, helpdesks etc.5

Choose one or moremonitoring approachesCalculation-based(chapter 3)Measurement-based(chapter 4)Fall-back(chapter 5)Activity data (3.1)Operator’s control (3.1.1)• Route CO-1/2a/2b/3EN 14181, EN 15259 orother standardsUncertainty over the wholeinstallation (also see Annex///, section 8.4)Not operator’s control (3.1.2)• Route CT-1/2Calculation factors (3.2)• “1/3” rule• Reference to GD5 “Sampling& Analysis”Figure 2:Relevant chapters and sections in this document regarding determination ofuncertaintyThis document is divided into chapters according to the monitoring approach applied: Calculation-based approaches are discussed in chapter 3; For measurement-based approaches, see chapter 4; Fall-back approaches are described in chapter 5.Due to the availability of various simplification options under the MRR, there are usuallyseveral routes by which an operator can demonstrate that the uncertainty levels correspondingto certain tiers are achieved, as shown in Figure 2. Those options (orroutes) are assigned codes throughout this document. For example, if a calculationbasedmethodology is applied and the activity data of a source stream are monitoredby a measurement system outside the operator’s own control, chapter 3 and sections3.1 and 3.1.2 (Route CT-1, CT-2 or CT-3) in particular will provide relevant guidancefor assessing uncertainty related to that activity data.9

3 UNCERTAINTY FOR CALCULATION-BASEDAPPROACHESThe following formula shows the calculation of emissions related to the most commoncase, i.e. the combustion of fuels, using the standard calculation method in accordancewith Article 24(1):Example: Calculation-based monitoring of combustions of fuelsWhere:Em ....... Emissions [t CO 2 ]Em = AD ⋅ NCV ⋅ EF ⋅OF⋅( 1−BF)AD ....... Activity data (= fuel quantity) [t or Nm 3 ]NCV..... Net Calorific Value [TJ/t or TJ/Nm 3 ]EF ........ Emission factor [t CO 2 /TJ, t CO 2 /t or t CO 2 /Nm 3 ]OF ....... Oxidation factor [dimensionless]BF ........ Biomass fraction [dimensionless]For each parameter the MRR defines tiers that shall apply, subject to being technicallyfeasible and not incurring unreasonable costs.Those parameters can be divided into the following two types:Activity data (AD): Tiers here relate to the required minimum uncertainty over thereporting period of the amount of fuel burned (uncertainty is discussed in section3.1 for this purpose).Calculation factors (NCV, EF, Carbon Content,…): Tiers here relate to specificmethodology set out in the MRR for the determination of each factor, e.g. using defaultvalues or carrying out analyses (corresponding uncertainty issues are discussedin section 3.2).3.1 Activity dataPlease note that everything said here for the activity data of a source stream monitoredby a calculation-based approach is also applicable to the input or output materialof a source stream monitored by a mass balance approach.The tiers for activity data of a source stream (see section 4.5 of GD 1) are defined usingthresholds for a maximum uncertainty allowed for the determination of the quantityof fuel or material over a reporting period. Whether a tier is met, must be demonstratedby submitting an uncertainty assessment to the competent authority together withthe monitoring plan, except in case of installations with low emissions. For illustration,Table 1 shows the tier definitions for combustion of fuels. A full list of the tier thresholdsof the MRR is given in section 1 of Annex II of the MRR.10

Table 1:Tier No.Typical definitions of tiers for activity data based on uncertainty, given for thecombustion of fuels (for example).Definition1 Amount of fuel [t] or [Nm 3 ] over the reporting period 9 is determined with amaximum uncertainty of less than ± 7.5 %.2 Amount of fuel [t] or [Nm 3 ] over the reporting period is determined with amaximum uncertainty of less than ± 5.0 %.3 Amount of fuel [t] or [Nm 3 ] over the reporting period is determined with amaximum uncertainty of less than ± 2.5 %.4 Amount of fuel [t] or [Nm 3 ] over the reporting period is determined with amaximum uncertainty of less than ± 1.5 %.Note that the uncertainty is meant to refer to “all sources of uncertainty, including uncertaintyof instruments, of calibration, any additional uncertainty connected to how themeasuring instruments are used in practice, and of environmental impacts”, unlesssome simplifications are applicable. The impact of the determination of stock changesat the beginning and end of the period has to be included, where applicable (see examplein section 8.3 of Annex III).In principle there are two possibilities for determining the activity data in accordancewith Article 27(1):Based on continual metering of the process which causes the emissionsBased on aggregation of metered amounts separately delivered taking into accountrelevant stock changes.The MRR does not require every operator to equip the installation with measuring instrumentsat any cost. That would contradict the MRR’s approach regarding cost effectiveness.Instruments may be used which are eitherUnder the operator’s own control (see section 3.1.1), orUnder the control of other parties (in particular fuel suppliers; see section3.1.2). In the context of commercial transactions such as fuel purchase it is oftenthe case that the metering is done by only one of the trade partners. The other partnermay assume that the uncertainty associated with the measurement is reasonablylow, where such measurements are governed by legal metrological control. Alternatively,requirements on quality assurance for instruments, including maintenanceand calibration can be included in the purchase contracts. However, the operatormust seek a confirmation on the uncertainty applicable for such meters in orderto assess if the required tier can be met.Thus, the operator may choose whether to use his own instruments or to rely on instrumentsused by the supplier. However, a slight preference is given by the MRR tothe operator’s own instruments: If the operator decides to use other instruments despitehaving his own instruments at his disposal, he has to provide evidence to thecompetent authority that the supplier’s instruments allow compliance with at least thesame tier, give more reliable results and are less prone to control risks than the methodologybased on his own instruments. This evidence must be accompanied with asimplified uncertainty assessment.9 Reporting period is the calendar year.11

smallAn exception to this concerns Article 47(4) 10 which allows operators of installationswith low emissions to determine the amount of fuel or material by using available anddocumented purchasing records and estimated stock without comparing the quality oftheir own instruments with the suppliers’ instruments.Throughout this document, different ways of assessing uncertainty are discussed. Itshould be kept in mind that many of these options should be seen as simplifications ofthe complete uncertainty assessment. However, none of the simplified routes shouldbe considered as a preferred route. Generally the operator is always allowed to performan individual (complete) uncertainty assessment (see Annex III of this document).3.1.1 Measuring system under the operator’s own control3.1.1.1 General aspectssmallIf the operator uses metering results based on measuring systems under his own control,he has to ensure that the uncertainty threshold of the relevant tier level is met.Consequently an uncertainty assessment is necessary. Although operators of installationswith low emissions are exempt from the requirement to provide the uncertaintyassessment to the competent authority, they may still require such assessment fortheir own purposes, for example, to claim compliance with a particular activity data tier.There are several sources of uncertainty, in particular errors which are caused by alack of precision (in principle this is the meter’s uncertainty as specified by the manufacturerfor use in an appropriate environment, and certain conditions for its installation,such as length of straight piping before and after a flow meter) and a lack of accuracy(e.g. caused by aging or corrosion of the instrument, which may result in drift).Therefore the MRR calls for the uncertainty assessment to take account of the measuringinstrument’s uncertainty, as well as the influence from calibration and all otherpossible influencing parameters. However, in practice such uncertainty assessmentcan be demanding, and may sometimes exceed the resources of operators. For theambitious researcher, an uncertainty assessment “never ends”. It is always possible toconsider even more sources of uncertainty. Thus, there is a need for pragmatism andto focus on the most relevant parameters contributing to the uncertainty. The MRR allowsseveral pragmatic simplifications.Figure 3 shows different approaches for uncertainty assessment, laid down by theMRR to prove compliance with the tier requirements of the MRR.10 Article 47(4): “By way of derogation from Article 27, the operator of an installation with low emissions maydetermine the amount of fuel or material by using available and documented purchasing records and estimatedstock changes. The operator shall also be exempt from the requirement to provide the uncertaintyassessment referred to in Article 28(2) to the competent authority.”12

Measuring instrument issubject to national legalmetrological controlMeasuring instrument is notsubject to national legalmetrological controlMeasuring instrument isinstalled in an environmentappropriate for its usespecificationsRoute CO-1 Route CO-2a/2b Route CO-3Uncertainty = Maximumpermissible error in serviceallowed by national legalmetrological controlUncertainty = Maximumpermissible error specified forthat measuring instrument inserviceORUncertainty = Uncertaintyobtained by calibrationmultiplied by a conservativeadjustment factorSpecificuncertaintyassessmentFigure 3:Activity data for calculation-based approaches: Approaches for determination ofthe uncertainty achieved (“C” means calculation based, “O” means instrument isunder operator’s own control)The operator can simplify the uncertainty assessment, ifThe measuring instrument 11 is subject to legal metrological control (Route CO-1).In this case the maximum permissible error in service laid down in the relevant nationallegal metrological text can be used as the overall uncertainty.The measuring instrument 11 is not subject to national legal metrological control butis installed in an environment appropriate for its use specifications. Then the operatormay assume that the uncertainty over the whole reporting period as required bythe tier definitions for activity data in Annex II of the MRR equals:the maximum permissible error specified for that instrument in service(RouteCO-2a), orwhere available and lower, the uncertainty obtained by calibration, multiplied bya conservative adjustment factor for taking into account the effect of uncertaintyin service (Route CO-2b).Where those simplifications are not applicable, or do not show that the required tier ismet, a specific uncertainty assessment in accordance with Route CO-3 and Annex IIIneeds to be carried out. An operator is not obliged to use any of the simplified approaches.He can always use Route CO-3.11 Please note that the singular form "measuring instrument“ is used here for simplicity reasons. In the caseof more instruments being involved in the determination of the activity data of a single source stream thesimplifications apply to all of them. The uncertainty related to the resulting activity data in the units requiredcan be determined by error propagation (see Annex III).13

3.1.1.2 Selecting an approachAn operator looking for the simplest approach should first check if Route CO-1 is applicable,i.e. if the measuring instrument is subject to national legal metrological controland that at least the tier required 12 is met. If the maximum permissible error in servicelaid down in the relevant legislation for national legal metrological control is higherthan the uncertainty required for the tier to be met, the operator may use another, butless simplified approach, i.e. either Route CO-2a or CO-2b. Only if these do not leadto the required result would the operator have to carry out a specific uncertainty assessmentin accordance with Route CO-3 and Annex III.Whichever route is chosen, the result must be robust evidence that the uncertainty determinedmeets the tier required. Where this is not the case, the operator must takethe necessary steps to comply with the M&R Regulation by:Carrying out corrective action, i.e. installing a measurement system that meets thetier requirements, orproviding evidence that meeting the required tier is technically infeasible or wouldincur unreasonable costs, and using the next lower tier in accordance with the resultof the uncertainty assessment.3.1.1.3 Simplification (“Route CO-1”)Measuring instrument is subject to national legal metrological control (NLMC)Overall uncertainty = Maximum permissible error in service (NLMC)The first simplification allowed by the MRR is the most straightforward in practice:Where the operator demonstrates to the satisfaction of the CA, that a measuring instrumentis subject to national legal metrological control (NLMC), the maximum permissibleerror in service (MPES) allowed by the metrological control legislation may betaken as the overall uncertainty, without providing further evidence 13 . The most appropriateevidence for being under NLMC is a certificate of the official verification of theinstrument 14 .NLMC usually is applicable where market transactions (trades) require the referenceto accepted standards (traceability). Within NLMC each type of measuring instrumentis assessed by evaluating the measurement results obtained by a large number oftests.Generally, measuring instruments subject to NLMC are considered more reliable, becausean assessment of the measuring instrument is obligatory and the measuring in-12 For calculation based approaches Article 26 of the MRR defines which tier is to be applied, subject to installationcategory and source stream category. For more details see <strong>Guidance</strong> document No. 1.13 The philosophy behind this approach is that control is exerted here not by the CA responsible for the EUETS, but by another authority which is in charge of the metrological control issues. Thus, double regulationis avoided and administrative burden is reduced.14 Article 3(c) of the MID (2004/22/EC) defines: ‘legal metrological control’ means the control of the measurementtasks intended for the field of application of a measuring instrument, for reasons of public interest,public health, public safety, public order, protection of the environment, levying of taxes and duties,protection of the consumers and fair trading;14

strument is checked and calibrated (calibration, see Route CO-2b) by a governmentalauthority or by an entrusted accredited body.Background information on maximum permissible errors under NLMCUnder legal metrological control calibration is considered valid where the uncertainty resultingfrom the calibration procedure is lower than the maximum permissible error(MPE) in verification. The term “in verification” is a metrological term here and mustnot be confused with verification under the EU ETS.Moreover it is considered that the equipment under regular service is exposed to measurementconditions that might have impact on the measurement result. This aspect ledto the introduction of a parameter called the maximum permissible error in service(MPE in service = MPES). This value represents a fair estimation of the uncertainty of adevice under regular operation, which undergoes regular legal metrological controlcomplying with the associated regulations. It sets a threshold for simplified checkswhich could be applied during regular operation and has therefore to be considered asthe uncertainty which needs to be attributed to the daily operation of the measurementequipment. This means that the MPES is more appropriate for use to ensure a fair exchangeof goods, the ultimate objective of legal metrological control.For some measuring instruments the MPE “under rated operating conditions” 15 are regulatedin the Measuring Instruments Directive (2004/22/EC) (MID) or by the Non-Automatic Weighing Instruments Directive (2009/23/EC) (NAWI), which intends to createa common market for measuring instruments across EU Member States. MPE inservice is subject to national legislation. Metrological control systems usually apply afactor of 2 to convert the maximum permissible error derived in verification into the maximumpermissible error in service (MPES). It needs to be mentioned that this factor isnot derived from statistics (unlike the difference between standard and expanded uncertainty)but follows from general experience in legal metrology with measuring instrumentswhich have undergone successful type approval tests 16 .3.1.1.4 Simplification (“Route CO-2a”)Measuring instrument is not subject to national legal metrological control but isinstalled in an environment appropriate for its use specificationsOverall uncertainty = Maximum permissible error in serviceThe second simplification allowed by the MRR, applies to measuring instruments thatare not subject to national legal metrological control but are installed in an environmentappropriate for their use specifications.For the second ETS phase, the so-called ETSG guidance document 17 proposed asimplified approach, which allowed the overall uncertainty for a source stream’s activi-15 Annex I of the MID defines: “The rated operating conditions are the values for the measurand and influencequantities making up the normal working conditions of an instrument.“ Thus, the definition of theMPE provided in the MID refers to the MPE in service (MPES). However, it should be noted that theMID only regulates the placing on the market and putting into use. It does not regulate any calibrationor maintenance to be carried out in service.16 Resulting from specific experience for some kinds of appliances other values for this factor are commonlyused, ranging from 1.25 (e.g. for automatic weighing systems) up to 2.5 (e.g. for traffic speed meteringdevices).17 The notes are found as Annex at http://impel.eu/projects/emission-trading-proposals-for-futuredevelopment-of-the-eu-ets-phase-ii-beyond15

ty data to be approximated by the uncertainty known for a specific type of instrument,under the condition that other sources of uncertainty are sufficiently mitigated. This isconsidered to be the case in particular if the instrument is installed according to certainrequired conditions. The ETSG guidance note contains a list of instrument types andinstallation conditions which helps the user applying this approach.The MRR has picked up the principle of this approach and allows the operator to usethe “Maximum Permissible Error (MPE) in service” 18 (MPES) specified for the instrumentas overall uncertainty, provided that measuring instruments are installed in anenvironment appropriate for their use specifications. Where no information is availablefor the MPES, or where the operator can achieve better values than the default values,the uncertainty obtained by calibration may be used, multiplied by a conservativeadjustment factor for taking into account the higher uncertainty when the instrument is“in service”. The latter approach reflects Route CO-2b.The information source for the MPES 19 and the appropriate use specifications are notfurther specified by the MRR, leaving some room for flexibility. It may be assumed thatthe manufacturer’s specifications,specifications from legal metrological control, and guidance documents such as the Commission’s guidance 20are suitable sources for MPES. The uncertainties given there may only be taken asthe overall uncertainty, if the measuring instruments are installed in an environmentappropriate for their use specifications (including Steps 1 to 4 below are met). If this isthe case values taken from these sources can be considered as representing theMPES and no further corrections to that uncertainty value are necessary.The operator can assume he meets the MRR requirements in such cases, if he showsevidence that all of the requirements of the following four steps are met:18 The MPE in service is significantly higher than the MPE of the new instrument. The MPE in service is oftenexpressed as a factor times the MPE of the new instrument.19 Please note that MPE and MPES for instruments under NLMC are based on experience and they are nottransferable to industrial measurement. The same denomination for instruments not subject to NLMC isonly used for simplicity reasons.20 Annex II of this <strong>Guidance</strong> Document provides conservative values for uncertainty ranges of commonmeasuring instruments and additional conditions.16

Step 1: Operating conditions regarding relevant influencing parameters areavailable 21The manufacturer’s specification for that measuring instrument contains operatingconditions, i.e. description of the environment appropriate for its use specifications,regarding relevant influencing parameters (e.g. flow, temperature, pressure, mediumetc.) and maximum permissible deviations for these influencing parameters. Alternatively,the manufacturer may have declared that the measuring instrument complieswith an international standard (CEN or ISO standard) or other normative documents(such as recommendations by OIML 22 ), which lay down acceptable operating conditionsregarding relevant influencing parameters.Step 2: Operating conditions regarding relevant influencing parameters are metThe operator demonstrates evidence that the operating conditions regarding relevantinfluencing parameters are met. For this evidence operators should make a check-listof the relevant influencing parameters (for example, see section 8.1, in particular Table2 and Table 3) for different measuring instruments and compare for each parameterthe specified range with the used range. This list should be provided to the competentauthority as part of the uncertainty assessment when submitting a new or updatedmonitoring plan.The result for this step should be an assessment thatthe measurement instrument is installed appropriately,the measuring instrument is appropriate to measure the medium of interest,there are no other factors that could have adverse consequences on the uncertaintyof the measurement instrument.Only if all of this is the case, can it be assumed that the MPES provided in the suitablesource (see above) is appropriate for use without further correction.Step 3: Performing quality assured calibration proceduresThe operator shows evidence that the regular calibration (calibration, see Route CO-2b) is performed by an institute accredited in accordance with EN ISO/IEC 17025,employing CEN, ISO or national standards where appropriate. Alternatively, if calibrationis performed by a non-accredited institute or by a manufacturer's calibration, theoperator has to show evidence (e.g. by calibration certificate), of suitability and thatthe calibration is performed using the instrument manufacturer’s recommended procedureand the results comply with the manufacturer's specifications.21 ‘CE’ marked measuring instruments are in conformity with the essential requirements laid down in Annex Iof the MID. This Annex requires manufacturers to specify such appropriate operating conditions. If themanufacturer’s specifications do not contain requirements for operating conditions regarding relevant influencingparameters, the operator has to carry out an individual uncertainty assessment (Route CO-3).However, in simple cases, expert judgement might be sufficient, in particular for minor and de-minimissource streams and for installations with low emissions.22 Documents containing technical specifications adopted by the Organisation Internationale de MétrologieLégale (OIML). http://www.oiml.org/17

Step 4: Further quality assurance procedures for measuring activity dataUnder Article 58(3), the operator is required to establish, document, implement andmaintain various written procedures to ensure an effective control system, including inrelation to quality assurance of relevant measurement equipment, and handling of resultingdata. Where certified quality or environmental management systems are inplace 23 , e.g. EN ISO 9001, EN ISO 14001, EMAS, to ensure that control activities (calibration,maintenance, surveillance and loss/failure management etc.) are carried out,it is recommended that these systems also include the quality assurance for measuringactivity data under the EU ETS.Unless all of the requirements of the four steps are fulfilled it cannot be assumed thatthe MPES taken from suitable sources (see above) can be used for uncertainty withoutfurther corrections. However, overall uncertainties might be calculated by combinationof the uncertainties provided in the suitable sources and a conservative estimateof the uncertainty related to the parameters causing this non-compliance, e.g.flow rate is partially outside the normal operating range, by the means of error propagation(see Route CO-3 and Annex III).3.1.1.5 Simplification (“Route CO-2b”)Measuring instrument is not subject to national legal metrological control but isinstalled in an environment appropriate for its use specificationsOverall uncertainty=Uncertainty from calibration × conservative adjustment factorCalibration 24The performance of regular calibration is the process where metrology is applied tomeasurement equipment and processes to ensure conformity of measuring instrumentsin use with a known international measurement standard. This is achieved byusing calibration materials or methods that ensure a closed chain of traceability to the“true value” performed as a measurement standard.Calibration should, if possible, be carried out by an accredited laboratory. Appropriatecalibration procedures and intervals may be found in the manufacturer’s specification,standards provided by accredited laboratories, etc. 2523 A control system is usually established in the installation for other purposes such as quality control or minimizingcosts. In a lot of cases material and energy flows are also of special relevance for other internalreporting systems (such as financial controlling).24Also see “EA 4/02 - <strong>Guidance</strong> to Expression of Uncertainty of Measurement in Calibrationhttp://www.europeanaccreditation.org/Docs/0002_Application%20documents/0002_Application%20documents%20for%20Laboratories%20Series%204/00100_EA-4-02rev01.PDF18

Example: Requirements for calibration of a flow meter for nonaqueousliquids with static start/stop measurementFor calibration the following aspects need to be considered:The flow meter is installed in accordance with the manufacturer’s specifications.The flow meter as well as the rest of the whole calibration system arefilled completely and are free from gases.The flow meter is at operating temperature.All parameter settings, to the extent available, should be documented.During zero flow rate before and after the measurement no signal indicatinga flow is detected.The calibration conditions (flow rate, temperature, pressure, liquidtype,…) are within the operating conditions.The flow rate is stable. The pressure must be high enough to avoid gasification or cavitation 26 .Density and viscosity have an influence on the calibration curve as well.Therefore, it is optimal to calibrate under the same conditions as during(intended) normal operation and to use the same, if available, or similarliquids.Adjusting to zero (“zeroing”) is to be done before and not during ameasurement series. Conditions of the liquid (temperature, pressure)are to be documented at the moment of zeroing. Zeroing is not requiredif the output signal for zero flow rate is lower than the range for the zerovalue provided by the manufacturer.The core element of each calibration procedure is the comparison of measurement resultswith a reference standard by applying a procedure which enables the determinationof a calibration function and of measurement uncertainties. The result of calibrationwill be a reliable assessment of the calibration function, its linearity (where this isa requirement) and the measurement uncertainty. The uncertainty obtained by calibrationshould, to the extent possible, relate to the operating range of the measuring instrumentin actual use. Thus, the calibration procedure should reflect to the extentpossible the operating conditions where the instrument is installed (i.e. where it is actuallyapplied).In many cases the measurand of interest is not measured directly but rather calculatedfrom other input quantities with a functional relationship, e.g. a volumetric flow (f V ) iscalculated by measuring inputs like density (ρ) and pressure difference (∆p) throughthe relationship f V =f V (ρ, ∆p). The uncertainty related to the measurand of interest will25 Also see “International vocabulary of metrology”http://www.bipm.org/utils/common/documents/jcgm/JCGM_200_2008.pdfNOTE 1 A calibration may be expressed by a statement, calibration function, calibration diagram, calibrationcurve, or calibration table. In some cases, it may consist of an additive or multiplicative correction ofthe indication with associated measurement uncertainty.NOTE 2 Calibration should not be confused with adjustment of a measuring system, often mistakenlycalled “self-calibration”, nor with [metrological] verification of calibration.26 Cavitation is the formation and then immediate implosion of cavities in a liquid, which may occur when aliquid is subjected to rapid changes of pressure, e.g. in turbines.19

then be determined as the combined standard uncertainty via error propagation 27 (seeAnnex III). For the combined standard uncertainty associated with the measurementresult, uncertainty contributions of long term drift and operational conditions are alsoimportant influences which have to be considered (besides the uncertainty associatedwith calibration).The expanded uncertainty of measurement is achieved by multiplying the combinedstandard uncertainty by a coverage factor. This factor is often taken to equal 2 fornormal distributions of data (Gaussian distributions). A factor of 2 corresponds to aprobability of 95% that the correct value is covered (i.e. a 95% confidence interval).Note that this coverage factor is still part of the expression of the uncertainty of measurementin calibration. This coverage factor is not the conservative adjustment factor(see below).Frequencies of calibrationDepending on the type of measuring instrument and the environmental conditions theuncertainty of a measurement might increase over time (drift). To quantify and to mitigatethe increase of uncertainty resulting from drift an appropriate time interval for recalibrationis necessary.In the case of a measuring instrument subject to NLMC (Route CO-1) the frequency ofcalibration (re-calibration) is regulated by the relevant legal text.For other measuring instruments intervals for re-calibration should be determined onthe basis of information provided by e.g. manufacturer’s specifications or other suitablesources. As the result of every calibration allowing quantification of the drift thathas occurred, time series analysis of previous calibrations may also be helpful to determinethe relevant calibration interval. Based on this information the operator shoulduse appropriate calibration intervals subject to the CA’s approval.In any case the operator has to check annually if the measuring instruments used stillcomply with the tier required (under point (b) of the first paragraph of Article 28).Industry practiceVarious situations need to be guarded against when it comes to calibration in industrialcircumstances, includingsimplifications for particular applications that do not then meet requirements forcalibration according to legal standards;single-point-tests or short checks that may be designed, for example, for checkingthe zero value and for providing day to day quality assurance, but which do notconstitute full calibration;postponement of calibrations due to favourable ad-hoc checks (suggesting properoperation of monitoring equipment) and due to the costs involved;failure to follow-up the results of the calibration by making adequate corrections.Moreover, a problem may occur when a device is not easily accessible for calibration,e.g. it cannot be de-installed for checks or calibration during operation of the installationand the process cannot be shut-down without major disruption to the installation27 It is more appropriate to call it “propagation of uncertainty” although “error propagation” is more frequentlyused.20

or to the security of supply associated with the product. There may be long periods betweenshut-downs of the production process and in such cases a periodic calibrationaccording to shorter intervals may not be feasible.Where only limited possibilities for calibration exist, the operator must seek approvalby the CA for an alternative approach, enclosing in the submission of the monitoringplan any relevant evidence with regards to technical feasibility or unreasonablecosts 28 . The hierarchy 29 of Article 32(1) for application of different standards should beconsidered.Conservative adjustment factorTo take into account any further random as well as systematic errors in service, theuncertainty obtained from calibration (expanded uncertainty, see above) is to be multipliedby a conservative adjustment factor. The operator should determine this conservativeadjustment factor, e.g. based on experience, subject to the approval of theCA. In the absence of any information or experience the use of a harmonised factor of2 is recommended as pragmatic yet appropriate approach. The result obtained maybe used as the overall uncertainty without further corrections.A conservative adjustment factor is only applicable if the measuring instrument is usedwithin the use specifications in accordance with Article 28(2), last sub-paragraph.Consequently, the requirements described for Route CO-2a (step 1 to step 4) have tobe met. If those requirements are not met this simplification route is not applicable andspecific uncertainty assessment described under Route CO-3 and Annex III is required.3.1.1.6 Full uncertainty assessment (“Route CO-3”)Full uncertainty assessment (“Route CO-3”)The operator is always entitled to carry out a specific uncertainty assessment, e.g. ifthe operator is of the opinion that this provides more reliable results. If this is the caseor where none of the simplification routes (Routes CO-1 or CO-2a/2b) are possible, anuncertainty assessment in accordance with Annex III has to be carried out.It is important to note that the obligation to carry out a specific uncertainty assessmentdoes not necessarily mean that this assessment has to be completely started fromnew. In many cases some prerequisites may apply concerning the simplificationsRoutes CO-1 or CO-2a/2b. In these cases uncertainties gathered from there might bestarting points for further calculations, e.g. via error propagation (see Annex III, in par-28 MRR Article 59(1), 2 nd sub-paragraph requires: “Where components of the measuring systems cannot becalibrated, the operator or aircraft operator shall identify those in the monitoring plan and propose alternativecontrol activities.”29 Article 32(1): “The operator shall ensure that any analyses, sampling, calibrations and validations for thedetermination of calculation factors are carried out by applying methods based on corresponding ENstandards. Where such standards are not available, the methods shall be based on suitable ISO standardsor national standards. Where no applicable published standards exist, suitable draft standards, industrybest practice guidelines or other scientifically proven methodologies shall be used, limiting samplingand measurement bias.”21

ticular section 8.2). This approach not only presents a more pragmatic and less burdensomeway for operators to assess uncertainty it may in most cases also providemore reliable results.Example: An operator is using a turbine meter subject to national legal metrologicalcontrol for the consumption of a liquid source stream. As the MRR requiresconverting the volumetric flow into mass flows the operator has to determinethe density of the liquid. As this is determined regularly by an aerometerno simplification, i.e. Route CO-1 or Route CO-2a/2b applies for the sourcestream if expressed in tonnes. However, the operator may be well advised touse the uncertainty laid down in the relevant national legal metrological text relatedto the determination of the volume in the overall uncertainty calculation byerror propagation (see section 8.3, in particular example 7).3.1.2 Measuring system not under the operator’s own control3.1.2.1 General aspectsThe operator may use a measurement system outside his own control to determineactivity data, provided that this system complies with at least as high a tier, gives morereliable results and is less prone to control risks 30 than using his own instruments, ifavailable. For these cases activity data may be determined either byamounts taken from invoices issued by the trading partner, orusing direct readings from the measurement system.Whichever approach is used, the same tiers for activity data are required as for systemsunder the operator’s own control (see section 3.1.1). The only difference is howthe operator can demonstrate this compliance and what simplifications may be applied.In the case of invoices providing the primary data for determining the material or fuelquantity, the MRR requires the operator to demonstrate that the trade partners are independent.In principle, this should be considered a safeguard for ensuring that meaningfulinvoices exist. In many cases it will also be an indicator of whether national legalmetrological control (section 3.1.1, Route CO-1) is applicable.Note that there is a “hybrid” possibility allowed by the M&R Regulation: The instrumentis outside the control of the operator (section 3.1.2), but the reading for monitoring istaken by the operator. In such a case the owner of the instrument is responsible formaintenance, calibration and adjustment of the instrument, and ultimately for the applicableuncertainty value, but the data on fuel or material quantity can be directlychecked by the operator. This is a situation frequently found for natural gas meters.Figure 4 shows the way provided by the MRR to comply with the tier requirements incase of measurement systems not under the operator’s control.30 For guidance on risk assessment see <strong>Guidance</strong> document No. 6 (Data flow and control activities).22

• Use amounts from invoices, provided that acommercial transaction between twoindependent trade partners takes place• Use of direct readings from themeasurement systemMeasuring instrument issubject to national legalmetrological controlMeasuring instrument is notsubject to national legalmetrological controlrequirements under nationallegal metrological control are atleast as stringent as therequired tierRoute CT-1requirements under nationallegal metrological control areless stringent than therequired tierRoute CT-2Route CT-3Uncertainty = Maximumpermissible error in serviceallowed by national legalmetrological controlObtain evidence on the applicable uncertainty fromthe trade partnerFigure 4:Activity data for calculation-based approaches: Approaches for determination ofthe uncertainty achieved (“C” means calculation-based, “T” means instrument isunder trading partner control)The operator can simplify the uncertainty assessment:If the measuring instrument is subject to legal metrological control, the maximumpermissible error laid down in the relevant national legal metrological text can beused as the overall uncertainty for assessing whether the tier requirements in accordancewith Article 26 are met (Route CT-1).If the applicable requirements under national legal metrological control are lessstringent than the uncertainty threshold of the tier required in accordance with Article26, the operator may obtain evidence from the trade partner concerning the uncertaintythat is actually applicable (Route CT-2).If the measuring instrument is not subject to national legal metrological control, theoperator may obtain evidence from the trade partner relating to the uncertainty concerned(Route CT-3).As discussed in section 3.1.1.2, the operator must ensure that the required tier in accordancewith Article 26 can be achieved. If not, either corrective action is required ora lower tier may be applied where evidence for unreasonable costs or technical infeasibilitycan be provided (as long as this still complies with at least as high a tier, givesmore reliable results and is less prone to control risks than the use of available instrumentsunder the operator's own control).3.1.2.2 Simplification (“Route CT-1”)Measuring instrument of the trade partner is subject to national legal metrologicalcontrol (NLMC).23

Overall uncertainty = Maximum permissible error in service (MPES)This simplification is applicable for the same reasons and under the same conditionsas given under section 3.1.1.3, Route CO-1. The operator must still be able to demonstratethat the trade partner's measuring instrument complies with at least as high a tieras an instrument available under the operator's own control and gives more reliableresults and is less prone to control risks.3.1.2.3 “Route CT-2”The operator shall obtain evidence of the applicable uncertainty from the tradepartner responsible for the measurement system.If the applicable requirements under national legal metrological control are less stringentthan the tier requirements of Article 26, the operator has to obtain evidence fromthe trading partner that the required tiers are met. The operator must be able todemonstrate that the trade partner's measuring instrument complies with at least ashigh a tier as an instrument available under the operator's own control and gives morereliable results and is less prone to control risks.This may also be based on an uncertainty assessment as explained in Annex III, usinginformation on the measuring instruments obtained from the trade partner. Please alsosee the information given under Route CO-3 (section 3.1.1.6).3.1.2.4 "Route CT-3”The operator shall obtain evidence of the applicable uncertainty from the tradepartner responsible for the measurement system.This route is similar to Route CT-2 above. In such a case where the transaction is notsubject to NLMC, the operator has to obtain evidence from the trading partner that therequired tiers of Article 26 are met. The operator must be able to demonstrate that thetrade partner's measuring instrument complies with at least as high a tier as an instrumentavailable under the operator's own control and gives more reliable resultsand is less prone to control risks.This may also be based on an uncertainty assessment as explained in Annex III, usinginformation on the measuring instruments obtained from the trade partner. Please alsosee the information given under Route CO-3 (section 3.1.1.6).24

3.2 Calculation factorsIn contrast to the tiers for activity data, the tiers for calculation factors 31 are not basedon uncertainty thresholds being met, but instead determinations involving default valuesor values derived from laboratory analyses. However, determinations involving laboratoryanalyses are linked to required frequencies for analyses (Article 35), and oneoption allowed for determining the required frequency is expressed in terms of the“uncertainty” related to the frequency of analyses. Article 35(2) states:“The competent authority may allow the operator to use a different frequency thanthose referred to in paragraph 1, where minimum frequencies are not available orwhere the operator demonstrates one of the following:a) based on historical data, including analytical values for the respective fuels ormaterials in the reporting period immediately preceding the current reportingperiod, any variation in the analytical values for the respective fuel or materialdoes not exceed 1/3 of the uncertainty value to which the operator has toadhere with regard to the activity data determination of the relevant fuel ormaterial…“It should be noted that the uncertainty assessment required in this case is differentand the detail is not considered within the scope of this document. Instead, the topic iscovered more specifically by <strong>Guidance</strong> document No. 5: "<strong>Guidance</strong> on Sampling &Analysis" (see section 1.3).31 The MRR defines in Article 3(7): ‘calculation factors’ means net calorific value, emission factor, preliminaryemission factor, oxidation factor, conversion factor, carbon content or biomass fraction25

4 UNCERTAINTY FOR MEASUREMENT-BASEDAPPROACHESFor a measurement-based approach including monitoring of N 2 O, Annex I of the MRRrequires a list of all relevant equipment, indicating its measurement frequency, operatingrange and uncertainty. The MRR does not mention any circumstances underwhich simplifications to determine the uncertainty apply, as there are for calculationbasedapproaches.However, Article 42 requires that all measurements shall be carried out based on thefollowing standards:EN 14181 Stationary source emissions – Quality assurance of automated measuringsystems,EN 15259 Air quality – Measurement of stationary source emissions – Requirementsfor measurement sections and sites and for the measurement objective, planand reportAnd other corresponding EN standards.EN 14181 for example contains information about quality assurance procedures (QAL2 and 3) to minimise the uncertainty as well as guidelines on how to determine the uncertaintyitself. For QAL 1 guidance can be found in EN ISO 14956 Air quality - Evaluationof the suitability of a measurement procedure by comparison with a requiredmeasurement uncertainty.Article 42 further states: “Where such standards are not available, the methods shallbe based on suitable ISO standards, standards published by the Commission or nationalstandards. Where no applicable published standards exist, suitable draft standards,industry best practice guidelines or other scientifically proven methodologiesshall be used, limiting sampling and measurement bias.The operator shall consider all relevant aspects of the continuous measurement system,including the location of the equipment, calibration, measurement, quality assuranceand quality control.”In the case that suitable standards or guidelines do not contain information about thedetermination of the uncertainty, some aspects for this determination may be takenfrom Annex III.26

5 UNCERTAINTY FOR FALL-BACK APPROACHESAn operator may apply a fall-back methodology, i.e. a monitoring methodology notbased on tiers, for selected source streams or emission sources, provided that all ofthe following conditions are met:applying at least tier 1 under the calculation-based methodology for one or moremajor source streams or minor source streams and a measurement-based methodologyfor at least one emission source related to the same source streams is technicallynot feasible or would incur unreasonable costs;the operator assesses and quantifies each year the uncertainties of all parametersused for the determination of the annual emissions in accordance with the ISOGuide to the Expression of Uncertainty in Measurement (JCGM 100:2008) 32 , or anotherequivalent internationally accepted standard, and includes the results in theannual emissions report;the operator demonstrates to the satisfaction of the competent authority that byapplying such a fall-back monitoring methodology, the overall uncertainty thresholdfor the annual level of greenhouse gas emissions for the whole installation does notexceed7.5% for category A installations,5.0% for category B installations and2.5% for category C installations.Further guidance for assessing the uncertainty can be found in Annex III, in particularin section 8.4.32 (JCGM 100:2008) Evaluation of measurement data – Guide to the expression of uncertainty in measurement(GUM): http://www.bipm.org/en/publications/guides/gum.html27

6 ANNEX I: ACRONYMS AND LEGISLATION6.1 Acronyms usedEU ETS ...... EU Emission Trading SchemeMRV ........... Monitoring, Reporting and VerificationMRG 2007 .. Monitoring and Reporting GuidelinesMRR ........... Monitoring and Reporting Regulation (M&R Regulation)MID............. Measurements Instruments Directive (MID 2004/22/EC)MP .............. Monitoring PlanCA ............. Competent AuthorityNLMC ......... National legal metrological controlETSG ......... ETS Support Group (a group of ETS experts under the umbrella of theIMPEL network, who have developed important guidance notes for theapplication of the MRG 2007)CEMS ......... Continuous Emission Measurement SystemMPE ........... Maximum Permissible Error (term usually used in national legal metrologicalcontrol)MPES ......... Maximum Permissible Error in service (term usually used in national legalmetrological control)MS .............. Member State(s)GUM ........... ISO Guide to the Expression of Uncertainty in Measurement (JCGM100:2008), downloadable fromhttp://www.bipm.org/en/publications/guides/gum.html.28

6.2 Legislative textsEU ETS Directive: Directive 2003/87/EC of the European Parliament and of theCouncil of 13 October 2003 establishing a scheme for greenhouse gas emission allowancetrading within the Community and amending Council Directive 96/61/EC,most recently amended by Directive 2009/29/EC. Download consolidated version:http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:2003L0087:20090625:EN:PDFM&R Regulation: Commission Regulation (EU) No. 601/2012 of 21 June 2012 on themonitoring and reporting of greenhouse gas emissions pursuant to Directive2003/87/EC of the European Parliament and of the Council.http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2012:181:0030:0104:EN:PDFA&V Regulation: Commission Regulation (EU) No. 600/2012 of 21 June 2012 on theverification of greenhouse gas emission reports and tonne-kilometre reports and theaccreditation of verifiers pursuant to Directive 2003/87/EC of the European Parliamentand of the Council.http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2012:181:0001:0029:EN:PDFMRG 2007: Commission Decision 2007/589/EC of 18 July 2007 establishing guidelinesfor the monitoring and reporting of greenhouse gas emissions pursuant to Directive2003/87/EC of the European Parliament and of the Council. The download ofthe consolidated version contains all amendments: MRG for N 2 O emitting activities,aviation activities; capture, transport in pipelines and geological storage of CO 2 , andfor the activities and greenhouse gases only included from 2013 onwards. Download:http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:2007D0589:20110921:EN:PDF29

7 ANNEX II: CONSERVATIVE MEASUREMENTUNCERTAINTIES FOR THE MOST COMMONMEASURING INSTRUMENTSThe following tables provide an overview of conservative measurement uncertaintiesfor certain categories of common measuring instrumentsThe uncertainty values and additional conditions presented in the tables belowshould be considered only if more specific information is not available from themanufacturer of the measuring instrument, or from normative documents suchas those published by OIML 33 . Also, these uncertainty values should be consideredonly if steps 1 to 4 (see section 3.1.1.4) are met. If this is not the caseRoute CO-2a is not applicable. For measuring instruments suitable for gasesand liquids relevant OIML documents are R137 and R117. For measuring instrumentsfor solids R76 is a suitable source.Please also note that a time period for recalibration is advised for each instrument.This implies that after each calibration the requirements to apply simplificationRoute CO-2b (section 3.1.1.5) might be applicable and provide more reliableresults. This option should always be considered before applying standardvalues listed below.Rotor meterMedium: gasRelevant standards: EN 12480:2002+A1:2006Uncertainty for 0-20% of the measurement range: 3%Uncertainty for 20-100% of the measurement range: 1.5%Conditions:- Once per 10 year cleaning, recalibration and if necessary adjusting- Annual inspection of the oil level of the carter- Application filter for polluted gas- Life span 25 yearsMedium: liquidUncertainty for 0-10% of the measurement range: 1%Uncertainty for 10-100% of the measurement range: 0.5%Conditions:- Once per 5 year cleaning, recalibration and if necessary adjusting(or at an earlier time when flow liquid of 3500 hours × maximumrange of the meter has run through the meter33 Documents containing technical specifications adopted by the Organisation Internationale de MétrologieLégale (OIML). http://www.oiml.org/30

- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 25 yearsTurbine meterMedium: gasRelevant standards: EN 12261:2002 + A1:2006Uncertainty for 0-20% of the measurement range: 3 %Uncertainty for 20-100% of the measurement range: 1,5%Conditions:- Once per 5 year cleaning, recalibration and if necessary adjusting- Annual visual inspection- Once per three months lubrication of bearings (not for permanentlubricated bearings)- Application filter for polluted gas- No pulsating gas stream- Life span 25 years- No overload of longer than 30 minutes › 120% of maximum measurementrangeMedium: liquidUncertainty for 10-100% of the measurement range: 0.5%Conditions:- Once per 5 year cleaning, recalibration and if necessary adjusting- Once per three months lubrication of bearings (not for permanentlubricated bearings)- Application filter for polluted liquid- Life span 25 years- No overload of longer than 30 minutes › 120% of maximum measurementrangeBellows meter / diaphragm meterMedium: gasRelevant standards: EN 1359:1998 + A1:2006Uncertainty for 0-20% of the measurement range: 7.5%Uncertainty for 20-100% of the measurement range: 4.5%Conditions:- Once per 10 year cleaning, recalibration and if necessary adjusting- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 25 years31

Orifice meterMedium: gas and liquidRelevant standards: EN ISO 5167Uncertainty for 20-100% of the measurement range: 3%Conditions:- Annual calibration of the pressure transmitter- Once per 5 years calibration of the orifice meter- Annual inspection of abrasion orifice and fouling- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 30 years- No corrosive gases and liquidsGuidelines for building in orifices, if not stated otherwise by the manufacturer:minimum of 50D free input flow length before the orifice and25D after the orifice: smooth surface of inner wall.Venturi meterMedium: gas and liquidRelevant standards: EN ISO 5167Gas: Uncertainty for 20-100% of the measurement range: 2%Liquid: Uncertainty for 20-100% of the measurement range: 1,5%Conditions:- Annual calibration of the pressure transmitter- Once per 5 years calibration of entire measuring instrument- Annual visual inspection- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 30 years- No corrosive gases and liquidsUltrasonic meterMedium: gas and liquidRelevant standards: ISO 17089-1:2010Gas: Uncertainty for 1-100% of the measurement range: 2%Gas (clamp on): Uncertainty for 1-100% of the measurement range:4%32

Liquid: Uncertainty for 1-100% of the maximum measurement range:3%Conditions:- Once per 5 years cleaning, recalibration and if necessary adjusting- Annual inspection of contact between transducer and tube wall.When there is not sufficient contact, the transducer assembly hasto be replaced according to the specifications of the manufacturer.- Annual inspection on corrosion of wall- Annual inspection of transducers- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 15 years- No disturbances in frequencies- Composition of medium is knownGuidelines for building in ultrasonic meters, if not stated otherwise bythe manufacturer: minimum of 10D free input flow length before themeter and 5D after the meterVortex meterMedium: gasGas: Uncertainty for 10-100% of the measurement range: 2,5%Liquid: Uncertainty for 10-100% of the measurement range: 2%Conditions:- Once per 5 years cleaning, recalibration and if necessary adjusting- Annual inspection of sensors- Annual inspection of bluff body- Annual inspection on corrosion of wall- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 10 years- Set-up is free of vibration- Avoid compressive shocksGuidelines for building in vortex meters, if not stated otherwise by themanufacturer: minimum of 15D free input flow length before the meterand 5D after the meterCoriolis meterMedium: gas and liquidGas: Uncertainty for 10-100% of the measurement range: 1,5%Liquid: Uncertainty for 10-100% of the measurement range: 1%33

Conditions:- Once per 3 years cleaning, recalibration and if necessary adjusting- Stress-free installation- Monthly control of adjusting zero point- Annual inspection of corrosion and abrasion- Annual check on sensors and transmitters- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 10 yearsOval gear meterMedium: liquidUncertainty for 5-100% of the measurement range: 1%Conditions:- Viscid liquids (oil): Once per 5 years cleaning, recalibration and ifnecessary adjusting- Thin liquids: Once per 2 years cleaning, recalibration and if necessaryadjusting- Annual inspection of abrasion- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 30 yearsElectronic Volume Conversion Instrument (EVCI)Medium: gasRelevant standards: EN 12405-1:2005+A1:2006Uncertainty for 0,95-11 bar and -10 – 40°C: 1%- Conditions: Once per 4 years recalibration and if necessary adjusting- Replace batteries (frequency is dependent on instructions manufacturer)- Annual maintenance according to instructions of manufacturer /general instructions measurement principle- Life span 10 years34

8 ANNEX III: FULL UNCERTAINTYASSESSMENT FOR SOURCE STREAMS8.1 IntroductionThis Annex should provide an overview about the general approach to assessuncertainties if no simplifications are applicable. For further details you mayconsult the GUM.In principle the uncertainty assessment shall comprisethe specified uncertainty of the applied measuring instrument,the uncertainty associated with the calibration andany additional uncertainties connected to how the measuring instrument isused in practice.If additional measurements such as pressure and temperature measurementare required, the uncertainty of these measurements has to be considered aswell. If the uncertainty information of the manufacturer cannot be applied, theoperator has to substantiate and justify that the deviations from the specificationdo not influence the uncertainty. If this is not possible, he has to make conservativeand substantiated estimations of the uncertainty. Possible influenceson the uncertainty include:Deviation from working rangeDifferent uncertainties subject to load or flow rateAtmospheric conditions (wind, temperature variation, humidity, corrodingsubstances)Operation conditions (adhesion, density and viscosity variation, irregularflow rate, in-homogeneity)Installation conditions (raising, bending, vibration, wave)Using the instrument for other medium than the one it is designed forCalibration intervalsLong-term stabilityThe general focus should be on the most significant parameters such as temperature,pressure (difference), flow rate, viscosity, etc., whichever applicable.Significant influences on the uncertainty have to be taken into account andevaluated. The uncertainty can be calculated with the appropriate error propagationformula. Some examples for the calculation of a specific uncertainty aregiven in this annex.Table 2 provides a list of various influencing parameters that might be relevantfor uncertainty assessment. It is not deemed complete, whereas in many casessome aspects can be neglected as they are likely to have minimal impact uponthe results. However, it could be used as first starting point when running a riskassessment with regard to the uncertainty of activity data and help focus on themost relevant influencing parameters. Table 3 provides some measuring instrumentspecific influencing parameters.35

Table 2:Influencing parameterrelated tothe equipmentand its installationInfluencing parameterrelated tothe medium beingmeasuredInfluencing parameters on the determination of activity dataGaseousStreamsSourceturbulences in gasstream impacts ofcladding temperatureof environmentlong-run behaviour(calibration andmaintenance frequency)acceptable measurementrangeelectromagneticfieldstemperaturepressurecompressibility factordew-point (for somegases only)corrosivenessFluidStreamsSourceturbulences in fluidstream, bubbling ofdissolved gasestemperature of environmentlong-run behaviour(calibration andmaintenance frequency)acceptable measurementrangeelectromagneticfieldsstorage capacityand monitoringphase changestemperaturedensityviscosityboiling or meltingpoint (for some rarecircumstances only)corrosivenessSolidStreamsSourceexposure to windand radiationtemperature of environmentlong-run behaviour(calibration andmaintenance frequency)position on scaleelectromagneticfieldsstorage capacities /volumesslope of conveyingbeltsstart and stop behaviouracceptable measurementrangestorage capacityand monitoringvibrationpurity / humidityaccessibility as netweight (e.g. packaging)handling of mediumimpacts by dryingdensityflow characteristics(e.g. related tograin size)adhesivenessmelting point (forsome rare constellationsonly)36

Table 3: Measuring instrument specific influencing parameters and way to validate/mitigatethemMetering of gases/liquidsMeasuring instrumentInfluencing parameterValidation/mitigation optionTurbine meter Intermittent flow, pulsation Appropriate operating parameters,avoid pulsation, e.g. by using controllinginstrumentsBellows meterCorrect detection of temperatureand pressureUse Electronic Volume ConversionInstrument EVCIOrifice meter, VenturimeterDamages, Roughness of the pipe,stability of pressure difference detectorsSatisfy EN ISO 5167 requirementsUltrasonic meter Strong noise signals Reduce noiseVortex meter Pulsation Avoid pulsationCoriolis meter Stress, vibration Build in compensatorsOval gear meter Resonances, pollution Dampers, filtersMetering of solidsMeasuring instrumentConveyor beltweighingInfluencing parameterAdhesion, sliding if belt is slantedValidation/mitigation optionUse horizontal beltWheel loader scale AdhesionZeroing after each measurementWagon weighbridgeWeighed object not fully on scale("full draught“)Use big enough scalesHopper weigher,truck weigher,crane weigherWindUse wind protection sites37

8.2 Error propagation lawsIn many cases the measurand of interest is not measured directly but rathercalculated from other input quantities being measured through a functional relationship,e.g. a volumetric flow (f V ) is calculated by measuring inputs like density(ρ) and pressure difference (∆p) through the relationship f V =f V (ρ, ∆p). The uncertaintyrelated to the measurand of interest will then be determined as the combinedstandard uncertainty via error propagation.For input quantities it is necessary to distinguish between:Uncorrelated (independent) input quantities, andCorrelated (interdependent) input quantities.If the operator uses different measuring instruments to determine the activitydata of parts of the source stream, the associated uncertainties can be assumedto be uncorrelated.Example: A gas flow measurement is converted from m³ to Nm³ by taking intoaccount temperature and pressure which are measured by separate measuringinstruments. These parameters can generally be considered as uncorrelated(see section 8.2.1).Example: The annual consumption of coal of a coal-fired power plant is determinedby weighing the batches delivered during the year with the same beltweigher. Due to drift-effects during operation in practice and due to uncertaintiesassociated to the calibration of the belt weigher, the uncertainties associatedwith the results of weighing are correlated (see section 8.2.2).However, this assumption has to be assessed carefully for each case asthere may be significant correlation between two input quantities if the samemeasuring instrument, physical measurement standard, or reference datumhaving a significant standard uncertainty is used.8.2.1 Uncorrelated input quantities:If uncorrelated input quantities X 1 ,..,X n are being used to calculate the measurandY=Y(X 1 ,..X n ) the uncertainty of Y can be determined by:UY=∂Y(∂X1⋅UX)2∂Y+ (∂X2⋅UX)2∂Y+ ... + (∂X1 2nn⋅UX)2(1)where:U Y ........ uncertainty (absolute value) of the measurand YU Xi ....... uncertainty (absolute value) of the input quantity X i38

Example 1: Uncorrelated input quantitiesY=Y(X 1 , X 2 ) is defined by the following relationship:Y = X 1 ⋅ X 2The partial derivatives are:∂Y∂X1= X2∂Y∂X2= X1The absolute uncertainty is then given by:UY=( X222 ⋅UX ) + ( X1⋅U)1 1X 2where:U Y ........ absolute uncertainty of measurand YU Xi ....... absolute uncertainty of input quantity X iThe relative uncertainty is given by:UYY= uY=( X2⋅UX1)X221+ ( X⋅ X221⋅UX2)2=U(XX11)2U+ (XX22)2=u2X1+ uX22where:u Y ........ relative uncertainty of measurand Yu Xi ........ relative uncertainty of input quantity X iThe square of the relative uncertainty of the measurand is therefore simply determinedas the sum of the squares of the relative uncertainties of the inputquantities.Example 2: Independent uncertainties of a sumA steam boiler for the production of process steam is operated by heating gasas fuel. The used heating gas is supplied to the boiler by ten different pipes.The amount of gas is determined by ten different standard orifice plates accordingto EN ISO 5167. The uncertainty associated with the determination of theannual consumption of heating gas (uncertainty of a sum) for the steam boiler iscalculated by following formula:2u ( U1)+ ( U 2 ) + ... + ( U= total x + x + ... + x1221010)2Where:u total ..... total (relative) uncertainty associated with the determination of the heatinggas39

U i ......... uncertainty (absolute value) of the individual standard orifice platesx i .......... quantities of heating gas that are measured annually by the different orificeplatesExample 3: Independent uncertainties of a productA combined heat and power plant with several boilers is fired by natural gas asthe only fuel. The annual quantity consumed is determined by a measurementsystem at the central transfer station (before distribution to the individual boilers)which consists of a turbine meter, a separate pressure measurement and aseparate temperature measurement. The turbine meter determines the flow rateat operating conditions.For emissions reporting the standard volume of natural gas is relevant. For theconversion of operating m³ into standard m³, measurements of pressure andtemperature have to be considered. Therefore the uncertainty associated withthe determination of the natural gas in standard m³ (uncertainty of a product) iscalculated by following formula:total2V2Tu = u + u + u2PWhere:u total ..... total (relative) uncertainty associated with the determination of naturalgasu V ........ (relative) uncertainty of the volume measurementu T ........ (relative) uncertainty of the temperature measurementu p ......... (relative) uncertainty of the pressure measurement8.2.2 Correlated input quantities:If correlated input quantities X 1 ,..,X n are being used to calculate the measurandY=Y(X 1 ,..X n ) the uncertainty of Y can be determined by:UY∂Y∂Y= ( ⋅UX) + ( ⋅U1X 2∂X∂X12∂Y) + ... + (∂Xn⋅UX n)(2)where:U Y ........ uncertainty (absolute value) of the measurand YU Xi ....... uncertainty (absolute value) of the input quantity X i40

Example 4: Correlated input quantitiesY=Y(X 1 , X 2 ) is defined by the following relationship:Y = X 1 ⋅ X 2If the example above was calculated for correlated input quantities the relativeuncertainty would be obtained as: 34 u Y = u X 1+ u X 2The relative uncertainty of the measurand is therefore simply determined as thesum of the relative uncertainties of the input quantities.Example 5: Correlated uncertainties of a sumA power plant is coal-fired. The annual consumption of coal is determined byweighing the batches delivered during the year with the same belt weigher. Dueto drift-effects during operation in practice and due to uncertainties associatedto the calibration of the belt weigher, the uncertainties associated with the resultsof weighing are correlated.Therefore, the uncertainty associated with the determination of the coal (uncertaintyof a sum) is calculated by following formula:utotalU1+ U2+ ... + U=x + x + ... + x12nnWhere:u total ..... total (relative) uncertainty associated with the determination of coalU i ......... uncertainty (absolute value) of the belt weigher (U 1 = U 2 = U n )x i ......... quantities of coal of the different batchesIn this case the (relative) uncertainty associated with the determination of coal isequal to the (relative) uncertainty of the belt weigher.Example 6: Correlated uncertainties of a productA mineral industry determines the loss on ignition by weighing the product on atable scale before and after the burning process. The loss on ignition is themass difference before and after the burning process related to the initialweight. The uncertainties associated with the results of the weighing are correlated,because the same table scale is used.Therefore, the uncertainty associated with the determination of the loss on igni-34 Please note that this is only applicable for the very special case where all of the input estimatesare correlated with correlation coefficients of 1. If the coefficient is different from 1, more complexfunctions for covariances are to be considered which are not within the scope of this document.For further reading please consult the GUM (see footnote 32.41

tion (uncertainty of a product) is calculated by the following formula:Where:u total= u 1+ u 2u total ..... is the total (relative) uncertainty associated with the determination of theloss on ignitionu 1,2 ....... (relative) uncertainty of the mass measurement before and after heating8.3 Case studiesExample 7: Uncertainty of the amount of stored fuelThe overall annual consumption of gasoil is calculated from the aggregated deliverieswith tank trucks. The trucks are equipped with a flow meter on the trucksubject to national legal metrological control with a maximum permissible errorof 0.5%. One truck is able to deliver 25,000 litres of gasoil. After the annualforecast the operator expects to require 750,000 litres annually on average overthe next year. Therefore, 30 tank truck deliveries per year are expected.The storage tank for gasoil at the installation has a capacity of 40,000 litres.With a cross section of 8m² the uncertainty of level reading is 2.5% of the totalcapacity.Note that the storage tank is capable of containing 40,000/750,000=5.3% of theannually used quantity and therefore has to be considered for the uncertaintyassessment 35 .The annual quantity Q of gasoil is determined by formula (10) in section 6.1.1 of<strong>Guidance</strong> Document 1:Where:Q = P − E +( S begin − SendP ......... Purchased quantity over the whole yearE ......... Exported quantity (e.g. fuel delivered to parts of the installation or otherinstallations which are not included in the EU ETS)S begin .... Stock of the gasoil tank at the beginning of the yearS end ...... Stock of the gasoil tank at the end of the year)As the quantity of purchased gasoil over the whole year (P) is not determinedby a single measurement but as the sum of many measurements, i.e. 30 truck35 According to Article 28(2), derogation is granted where the storage facilities are not capable ofcontaining more than 5% of the annual used quantity of the fuel or material considered. In suchcase the uncertainty of stock changes may be omitted from the uncertainty assessment.42

deliveries, P can be written as:P = P1 + P1+ .. + PWhere:P i ......... Purchased quantity from one truck30Now all input quantities for the determination of Q can be considered as uncorrelated36 . Under the assumption that no gasoil is being exported (E=0) the uncertaintycan therefore be determined in accordance with section 8.2.1 as anuncorrelated uncertainty of a sum:uQ=( US,begin)S2+ ( UbeginS,end− Send)2+ ( U12P1)+ P + .. + Pu Q ........ total (relative) uncertainty associated of Q+ .. + ( UU S, P ..... (absolute) uncertainty of the stock level reading or quantity provided byone tank30P30)2The uncertainty related to the stock level reading is the same for both readings.As the difference between S begin and S end cannot be predicted S begin -S end can beassumed as zero. If further all P i are considered as equal quantities havingequal absolute uncertainties the equation simplifies to:uQ=2 ⋅ ( U ) + n ⋅ ( US2PPi)2u Q=2 ⋅ (40000 ⋅ 2.5%)2+ 30 ⋅ (25000 ⋅ 0.5%)7500002= 0.21%As the activity data related to gasoil consumption has to be expressed in tonnesthe density of the fuel has to be taken into account. The uncertainty for determiningthe bulk density using representative samples is around 3%. Using theformula from section 8.2.1 for uncorrelated uncertainties of a product leads to:22u Q( tonnes)= uQ(Volume)+ udensity= 0,21% + 3% = 3.007%22Although the flow metering had a rather low uncertainty the conversion intotonnes displays that the influence of the uncertainty of the density determinationis the most significant contribution to the overall uncertainty. Future improvementsshould therefore relate to determination of the density with lower uncertainty.36 The level reading on the storage tank cannot be considered as being within one measurement seriesbecause of the long time period between the measurements (beginning and end of the year).However, as it is still the same measuring instrument that is being used, there may be some kindof correlation. Consideration as uncorrelated is an assumption for this particular example. In generalit has to be assessed, e.g. by determining correlation coefficients in accordance with theGUM 32 , if correlation really can be ruled out.43

Example 8: Uncertainty for source streams partly transferred toconnected installations not falling under the EU ETSWhen the installation is partly covered by EU ETS and not all parts of that installationfall under the scheme, the quantity measurement determined by an internalsub-meter (uncertainty is 5%) for the non-EU ETS part may have to besubtracted from the quantity of the source stream that is measured by the mainmeter which falls under national metrological control (uncertainty is 2%).Suppose the installation site uses 500,000 Nm³ natural gas annually. Out of thatamount of natural gas 100,000 Nm³ will be transferred and sold to an installationnot falling under EU ETS. To determine the consumption of natural gas ofthe EU ETS installation, the consumption of natural gas by that connected installationhas to be subtracted from the total natural gas consumption of the installationsite. To assess the uncertainty for the natural gas consumption of theEU ETS installation, following calculation is performed:u sourcestream=(2% ⋅500,000)2+ (5% ⋅100,000)500,000 + ( −100,000)2= 2.8%Please note, that the uncertainty of the main gas meter under national metrologicalcontrol does not have to be assessed. The uncertainty of the internal submeterthat is not guaranteed by national metrological control has to be assessedand confirmed before being able to determine the uncertainty associatedwith the source stream.8.4 Uncertainty over the whole installation (fall-backapproaches)This section is relevant if at least parts of the installation’s emissions aremonitored by a fall-back approach.Example 9: Overall uncertainty with a fall-back approachA category A installation has been exclusively burning natural gas during thesecond trading period with annual emission of 35,000 t CO 2 . As this fuel is obtainedby a commercial transaction subject to national legal metrological controlthe uncertainty related to the activity data may be 2.0% using the maximumpermissible error allowed by the relevant nation legislation. The 2.0% will alsobe the uncertainty related to the total emissions as all calculation factors appliedare default values are for reasons of simplicity not influencing uncertainty 37 .Due to the extension of the scope of the EU ETS from 2013 (third trading peri-37 Please note that also a default value (e.g. IPCC values or National Inventory values) exhibits anuncertainty related to that value. This uncertainty also has to be taken into account by calculatingthe uncertainty of the source stream from the independent uncertainties of the product (see example3) using error propagation.44

od) onwards an additional source stream will have to be included into the GHGpermit and therefore will be required to be monitored. The operator proves tothe satisfaction of the CA that applying at least tier 1, e.g. installing a measurementsystem, is technically not feasible and proposes to use a fall-back approach.The operator provides evidence in accordance with the GUM that anuncertainty assessment for that source stream gives an uncertainty (95 % confidenceinterval) of 18%. The expected emissions from that source stream are12,000 t CO 2 annually.When applying a fall-back approach for a category A installation the operatorhas to demonstrate that the uncertainty of the emissions for the whole installationdoes not exceed 7.5%. In the given example the operator has to calculatethe uncertainty using the equationEm = Em + EmtotalNGFBwhere:Em total .. total emissions of the installationEm NG ... emissions resulting from natural gas burning (35,000 t CO 2 )Em FB ... emissions resulting from the source stream monitored by a fall-backapproach (12,000 t CO 2 )As the (relative) uncertainty of the overall emissions can be interpreted as theuncertainties of a sum, the overall uncertainty is calculated by:u total=(2.0% ⋅35,000)235,000 + 12,000)+ (18% ⋅12,000)2= 4.8%The uncertainty related to the emissions over the whole installation is not exceeding7.5%. Therefore, the proposed fall-back approach is applicable.45